Preventing Youth Soccer Injury: A Randomized Cross-Over Trial By most accounts, soccer is the most popular sport in the world. An estimated 270 million people play or officiate organized soccer worldwide each year (Kunz, 2007), including over 3 million youth in the United States (US; http://www.usyouthsoccer.org/media_kit/keystatistics/). Almost half of participating children in the US are girls. Soccer is the 3rd most common cause of youth sports injury requiring emergency department visits in the US, the leading cause of sports concussion for girls, and among the top five leading causes of sports concussion for boys (Buzas et al., 2014; MMWR, 2011; Schwebel & Brezausek, 2014). We propose a simple but innovative strategy to reduce illegal and aggressive youth soccer play that may lead to injury: increase the number of referees regulating youth soccer matches from one to three. Our laboratory has conducted a series of studies that demonstrate children are more cautious in potentially dangerous situations when they recognize they are being watched more intensely and/or proximally by an adult authority figure. We believe this discovery, which has been repeatedly replicated in both laboratory and ecologically-valid field settings, would extend to the soccer field and that the presence of extra referees would reduce illegal and aggressive soccer play that leads to injuries, including concussions. To test this hypothesis, we propose a randomized cross-over trial. We will record approximately 300 recreational youth soccer games among teams of boys and girls ages 10 to 11, randomly assigning each game to have either one referee (current practice) or three referees (experimental intervention). We will use objective behavioral coding strategies to review recordings and tally multiple outcomes including six primary outcomes: injury incidents, injuries to the head, injuries requiring professional medical attention, player-to- player collisions, falls to the ground, and fouls. Primary data analysis will be conducted via linear mixed models examining difference in outcomes across games with 3 vs. 1 referee and accounting for potential game-related, referee-related, and temporal variables.